Blockchain Dynamic Sharding Model Based on Node Credibility

Blockchain Dynamic Sharding Model Based on Node Credibility

Introduction

Blockchain technology has emerged as a critical driver for digital transformation and the development of the metaverse. Governments worldwide, including China’s 14th Five-Year Plan, have recognized blockchain as a strategic digital industry to ensure secure data circulation and trusted sharing. Despite its potential, blockchain faces significant scalability challenges, including low transaction throughput, high latency, and substantial storage overhead. These limitations hinder its adoption in high-concurrency applications.

Among various scalability solutions, sharding has become a mainstream approach because it enhances performance without compromising decentralization. Traditional sharding models, such as Elastico, OmniLedger, and RapidChain, improve throughput by dividing the network into smaller, parallel-processing shards. However, these models introduce new challenges, including reduced security due to smaller shard sizes, malicious node aggregation, and load imbalance.

To address these issues, this paper proposes a Blockchain Dynamic Sharding Model Based on Node Credibility (CBDSM). The model introduces a node credibility assessment mechanism to evaluate nodes based on their hardware performance and consensus behavior, ensuring secure and efficient sharding. Additionally, a dynamic sharding algorithm balances node distribution and mitigates malicious attacks. Experimental results demonstrate that CBDSM improves throughput by 54%, reduces consensus failure rates by 46%, and decreases shard failure rates by 15.6% compared to existing models.

Related Work

Existing blockchain sharding solutions primarily focus on random node assignment to prevent malicious node concentration. For instance:
• Elastico and OmniLedger use randomness to distribute nodes across shards.

• RapidChain improves load balancing but still faces security risks due to smaller shard sizes.

• Monoxide relies on address-based sharding, which is predictable and vulnerable to targeted attacks.

• SkyChain and other reinforcement learning-based approaches dynamically adjust sharding parameters but introduce high computational overhead.

Recent studies have explored reputation-based sharding to enhance security:
• TBSD uses trust mechanisms and genetic algorithms to distribute malicious nodes.

• RepChain employs a dual-chain architecture (transaction chain + reputation chain) but increases system complexity.

• Cuckchain integrates a reputation-based node assignment but lacks dynamic adjustments.

Despite these advancements, existing models fail to comprehensively assess node heterogeneity, leading to inefficiencies in shard allocation. CBDSM addresses these gaps by integrating performance-based and behavior-based credibility metrics, ensuring balanced and secure sharding.

Node Credibility Assessment Mechanism

Credibility Calculation

CBDSM evaluates nodes based on two dimensions: performance credibility and behavior credibility.

1. Performance Credibility
   This metric assesses a node’s hardware capabilities, including:
   • Network bandwidth

• Storage capacity

• CPU performance

• Processing speed

Nodes with superior hardware receive higher credibility scores, ensuring efficient transaction processing.

2. Behavior Credibility
   This metric evaluates a node’s consensus participation and reliability, considering:
   • Consensus accuracy: Measures correct transaction validation rates.

• Response speed: Tracks the time taken to validate transactions.

Nodes that consistently submit accurate results and respond quickly gain higher behavior credibility.

3. Comprehensive Credibility
   A weighted combination of performance and behavior credibility determines a node’s overall trustworthiness. Nodes are then classified into three categories:
   • High-credibility nodes: Eligible for leadership roles and cross-shard transactions.

• Ordinary nodes: Participate in intra-shard consensus.

• Malicious nodes: Removed from the network to prevent attacks.

Node Classification Algorithm

To efficiently categorize nodes, CBDSM employs a threshold-based classification algorithm:

1. Nodes are evaluated at each epoch.
2. Their credibility scores are compared against predefined thresholds.
3. Malicious nodes are immediately excluded, while high-credibility nodes are prioritized for critical tasks.

This approach reduces computational complexity and ensures dynamic security adjustments.

Credibility Update Process

The credibility update mechanism operates in four phases:

1. Request Phase: Clients send credibility update requests to primary nodes.
2. Broadcast Phase: Primary nodes distribute credibility updates across shards.
3. Verification Phase: Nodes validate updates to ensure accuracy.
4. Update Phase: Nodes synchronize their credibility records.

This process ensures real-time credibility adjustments, maintaining system integrity.

Dynamic Sharding Model

Architecture Overview

CBDSM divides the blockchain network into multiple shards, each operating independently. Key components include:
• Transaction Pool: Transactions are hashed and assigned to specific shards.

• Intra-Shard Consensus: Each shard processes transactions using Practical Byzantine Fault Tolerance (PBFT).

• Cross-Shard Consensus: High-credibility nodes facilitate inter-shard transactions.

This architecture minimizes redundant storage and reduces network overhead.

Dynamic Sharding Process

The sharding lifecycle consists of four stages:

1. Initial Shard Construction
   • Nodes are assigned initial credibility scores.

• A balanced shard distribution is established using credibility-based allocation.

2. Consensus Phase
   • Nodes validate transactions via PBFT.

• Primary nodes record consensus behavior for credibility updates.

3. Credibility Evaluation
   • Nodes are reassessed at each epoch.

• Malicious nodes are removed, while high-credibility nodes are promoted.

4. Shard Reconfiguration
   • Underperforming shards are dynamically restructured.

• Nodes are redistributed to maintain load balance.

Sharding Algorithm

The dynamic sharding algorithm ensures secure and efficient node allocation:

1. Nodes are classified by credibility.
2. High-credibility and ordinary nodes are randomly distributed to prevent malicious clustering.
3. Shards are continuously monitored and reconfigured as needed.

This approach enhances security while minimizing reorganization costs.

Security and Performance Analysis

Resistance to Attacks

1. 1% Attack Mitigation
   • Random node distribution prevents malicious takeover.

   • Credibility thresholds ensure shards remain resilient.

2. Denial-of-Service (DoS) Protection
   • Invalid transactions are filtered before consensus.

3. Byzantine Fault Tolerance
   • Malicious nodes are quickly identified and removed.

   • The system tolerates up to 1/3 Byzantine nodes without failure.

Performance Evaluation

Experiments conducted on a Go-based testbed demonstrate CBDSM’s advantages:

1. Throughput
   • CBDSM achieves 6,225 TPS, outperforming OmniLedger (3,500 TPS) and RapidChain (3,774 TPS).

• Throughput scales linearly with shard count.

2. Storage Efficiency
   • Reduces storage overhead by 44% compared to non-sharded blockchains.

3. Consensus Reliability
   • Consensus failure rates drop by 46.7% compared to random sharding.

4. Load Balancing
   • Transaction distribution across shards is 10x more balanced than Monoxide.

Conclusion

CBDSM introduces a credibility-aware dynamic sharding model that enhances blockchain scalability and security. By integrating performance and behavior-based node assessment, the model prevents malicious node aggregation and ensures load balancing. Experimental results confirm significant improvements in throughput, failure resistance, and efficiency.

Future work will focus on optimizing cross-shard transactions and reducing consensus latency for broader adoption.

doi.org/10.19734/j.issn.1001-3695.2024.04.0243

Was this helpful?

YesNo

0 / 0